The Dynamics of Political Interest and News Media Avoidance: A Generational and Longitudinal Perspective

ABSTRACT

Despite the abundance of political information in today’s media environment, an increasing number of people tune out from the news. Scholars have traced these developments to both individual motivational factors gaining importance in the high-choice media environment and generational differences related to changing news habits and changing norms around news consumption. However, research that integrates these perspectives is scarce, and studies investigating this relationship longitudinally are especially rare. Thus, this paper investigates the conditional effects of (1) political interest and (2) age on the probability of avoiding traditional news media, and (3) whether these conditional effects have changed over time. Using data from 1986 to 2021, this study shows that the number of people tuning out from news media is increasing. The effect of political interest and age on the probability of news media avoidance has also increased over time. Additionally, there were only small age differences in the impact of political interest in earlier periods, but these differences have increased over time. Today, being uninterested in politics is clearly a stronger predictor of news media avoidance among younger people compared to older individuals.

KEYWORDS:

• News avoidance
• news consumption
• longitudinal study
• generational effects
• political interest
• quantitative audience research

Introduction

During recent decades, media systems in post-industrial Western democracies have transformed from low-choice to high-choice media environments (Van Aelst et al. 2017). Never have people been able to access so much information, with so little effort, in so little time. Despite this accessibility, it has not resulted in a universal increase in news media consumption. The trend in many countries is that news media use is going down (Blekesaune, Elvestad, and Aalberg 2012), and the number of people saying they actively avoid news is increasing (Newman et al. 2022). These developments could have democratic implications since news consumption is positively related to political knowledge and participation (Damstra et al. 2023; Ksiazek, Malthouse, and Webster 2010; Prior 2007).

This “paradox of high choice” (Edgerly 2022, 1829) has attracted much scholarly attention. One of the most prominent explanations of this paradox is rooted in the notion that the increase in the media supply forces people to become more selective in their media choices (Prior 2007). In turn, this growing preference-based media selection contributes to a news consumption gap, where people who are interested in news and politics continue to consume news while the uninterested increasingly tune out in favor of content more aligned with their tastes (Prior 2007). As a result, people with low interest in news are more likely to become news avoiders (Skovsgaard and Andersen 2020). Audience fragmentation along motivational lines might therefore exacerbate already existing inequalities in political involvement.

Related to these developments are increasing generational gaps in news consumption. Younger generations tend to consume less traditional media, relying instead to a greater degree on online media to get their news, with social media playing a prominent role in their news diets (Holt et al. 2013; Newman et al. 2022). These differences have been conceptualized as reflecting both changing news habits (Edgerly et al. 2018) and changing norms around news consumption among younger citizens in Western post-industrial democracies (Bennett 2008). While this implies that social and digital media may compensate for a generational decline in traditional news consumption (Karlsen, Beyer, and Steen-Johnsen 2020; Trilling and Schoenbach 2013), research suggests that people learn less from using these newer types of media (Amsalem and Zoizner 2023; Shehata and Strömbäck 2021). Thus, apart from preference-based fragmentation, understanding generational differences in news media consumption is important.

What is lacking in previous research are studies that integrate the motivational and generational perspectives during the transition from low-choice to high-choice media environments. Previous studies that have investigated the changing impact of motivational factors on news consumption over time (Strömbäck, Djerf-Pierre, and Shehata 2013) have not considered that this changing impact might look different across ages. Also, studies that have found that the impact of motivational factors differs across ages (Andersson 2019; Boulianne and Shehata 2022) lack a longitudinal perspective, making it hard to determine whether we are dealing with life-course effects (Quintelier 2007) or generational shifts in the relations toward news consumption (Thorson 2015). This lack of research into the relationships between changing audiences and a changing media environment is unfortunate since they are inextricably linked. Not only are they parallel processes but also, as Aalberg and colleagues (2013, 299) argued, when media choices become more individualized it opens a zone of heterogeneous socialization where the “transmission of singular cultural values from one generation to the next will get more difficult. Future studies need to explore whether news consumption gaps increase more among young citizens” (299).

Against this background, this study aims to investigate the conditional effects of (1) political interest and (2) age on the probability of avoiding news media, and (3) whether these conditional effects have changed over time. Empirically, this study uses repeated cross-sectional data (N = 141,973) measuring 35 years of news media use in Sweden.

Defining News (Media) Avoidance

Despite garnering substantial interest in recent decades, news avoidance research has been characterized by conceptual ambiguity with varying accounts of scope, causes and potential remedies (Skovsgaard and Andersen 2020). To remedy this, researchers must be clear about their definition of news avoidance, in particular concerning (1) whether definitional priority is given to intentionality or levels of news use and (2) what type of news are avoided.

Regarding the first, some studies give definitional priority to intentionality, defining and operationalizing news avoidance in terms of individuals’ deliberate actions or strategies (e.g. Toff and Kalogeropoulos 2020; Villi et al. 2022; Ytre-Arne and Moe 2021). This “intentional” (Skovsgaard and Andersen 2020) or “situational” (Palmer, Toff, and Nielsen 2023) view of news avoidance usually does not consider people’s actual news use but rather relies on self-reported measures of news avoidance behavior. Other research considers news avoidance as a characteristic of groups in society that habitually consume little to no news (Edgerly 2022; Ksiazek, Malthouse, and Webster 2010; Strömbäck, Djerf-Pierre, and Shehata 2013; Trilling and Schoenbach 2013). This perspective has been called “unintentional” (Skovsgaard and Andersen 2020) or “consistent” (Palmer, Toff, and Nielsen 2023) news avoidance. This article leans towards the latter perspective since we define news avoidance based on levels of news use. We do not discount the role of intentionality but regard it as having secondary significance in this study. After all, most people have—intentionally or unintentionally—occasionally avoided the news for some time, but this question is of secondary importance to understanding the dynamics behind a habitual estrangement from news, especially if this estrangement is tied to specific groups (Palmer, Toff, and Nielsen 2023).

The second distinction relates to which news is avoided. Research measuring traditional news media tends to find larger gaps in news consumption and higher levels of news avoiders (Strömbäck, Djerf-Pierre, and Shehata 2013) compared to studies analyzing more unspecified measures of online or social media news use (Gorski and Thomas 2021; Karlsen, Beyer, and Steen-Johnsen 2020), leading some to suggest that these might function as potential levelers of news consumption gaps (Boulianne 2011; Holt et al. 2013). However, research suggests that news consumption on these websites does not compensate for not using traditional media when learning about politics (Amsalem and Zoizner 2023; Shehata and Strömbäck 2021). Given these results, some scholars have argued that the broad categories of “online” and/or “social media” news use employed by some studies are too uninformative and scholars should be sensitive to what people actually do when they consume news online (Shehata and Strömbäck 2021).

Hence, this paper will focus on news media avoidance (i.e., legacy media outlets both online and offline), during the transition from a low- to a high-choice media environment. There are three reasons for this. First, most scholars agree that consumption of legacy media such as public service news and quality newspapers works as a leveler of political knowledge gaps among citizens, contributing to a more informed electorate with less partisan-selective exposure (Dahlgren 2019; Fraile and Iyengar 2014). Furthermore, political news exposure from traditional sources seems to have stronger mobilization effects compared to exposure on social media, especially for younger generations (Andersen et al. 2021). As such, to the extent that we consider habitual avoidance of news a topic of democratic concern (Palmer, Toff, and Nielsen 2023; Skovsgaard and Andersen 2020), these outlets are of certain importance, especially if we are witnessing increasing motivational and generational gaps. Second, on a broader scale news media audiences matter because “without them, the main purposes of journalism—acting as a watchdog, information source, intermediary between the people and government, and so forth—are meaningless” (Peters and Schrøder 2018, 1079). The decline of audiences shares has deteriorated the business models of legacy media, contributing to a crisis of journalism with increasing power concentration and decreasing news diversity in the contemporary media environment (Newman et al. 2022; Usher 2021; see Van Aelst et al. 2017 for review). The third reason concerns data availability. Because this study measures news avoidance over three decades, the measurements of news media need to be available for the whole period. As a result, native online media outlets and social media had to be excluded.

Given all this, the analytical focus of this article will be news media avoidance which we—adapting Skovsgaard & Andersen’s definition (2020, 463)—define as low news media consumption over a continuous period of time.

Individual Motivations and News Media Avoidance

Central to news audience research is the opportunity, motivation, and ability (OMA) framework, which states that any political behavior is contingent on three factors: opportunity, motivation, and ability (Luskin 1990). Regarding news consumption, opportunity refers to the availability of news media and media technology in the environment (Delli Carpini and Keeter 1996; Luskin 1990). Motivation refers to how motivated people are in seeking out news (Delli Carpini and Keeter 1996). Abilities encompass both physical attributes and cognitive abilities, such as the ability to process, contextualize, and store political information (Prior 2007).

The central argument of the OMA framework is that as opportunity structures have facilitated greater self-selection in the high-choice media environment, individual-level motivations will become stronger predictors of news consumption (Prior 2007). This is because the myriad of media available forces people to choose how to best spend their time and, consequently, preference-based choice will increasingly determine news use. In short, people uninterested in the news will increasingly tune out in favor of other types of media content, thereby becoming unintentional news avoiders (Skovsgaard and Andersen 2020).

Perhaps the most important individual-level motivation is political interest. Both single-country and comparative studies continuously show that the politically interested consume more news and are less likely to be news avoiders (Blekesaune, Elvestad, and Aalberg 2012; Gorski and Thomas 2021; Strömbäck, Djerf-Pierre, and Shehata 2013; Trilling and Schoenbach 2013). In sum, political interest is an important predictor of news use in the contemporary high-choice media environment.

Still, studies examining the claim that the impact of political interest has increased over time due to the proliferation of media choices have yielded mixed results. Some studies show that individual motivations are becoming more important in the high-choice media environment (Dahlgren 2019; Prior 2007; Strömbäck, Djerf-Pierre, and Shehata 2013). Other studies have found small, over-time changes in the importance of political interest (Aalberg, Blekesaune, and Elvestad 2013; Gorski and Thomas 2021; Karlsen, Beyer, and Steen-Johnsen 2020).

These differing results might be due to the choice of outlets measured. As previously discussed, some studies measure consumption of more traditional news outlets (Aalberg, Blekesaune, and Elvestad 2013; Dahlgren 2019; Strömbäck, Djerf-Pierre, and Shehata 2013), while others incorporate broader non-specific measures of generalized news use (Gorski and Thomas 2021; Karlsen, Beyer, and Steen-Johnsen 2020). Another potential explanation is whether one measures exposure to or avoidance of news. Higher political interest might be a strong predictor of higher total news consumption (Strömbäck, Djerf-Pierre, and Shehata 2013), although a modest or very small amount of political interest might be sufficient to stay tuned in (Dahlgren 2019, 302; Trilling and Schoenbach 2013). Thus, there is a need for longitudinal research on the changing impact of political interest regarding news avoidance. Following the OMA framework, we expect the following:

H1a: Having a low political interest increases the probability of avoiding news media; and

H1b: The size of the relationship between low political interest and probability of news media avoidance has increased over time.

Age, Generation, and News Media Avoidance

Age is one of the most important demographic factors predicting news media use (Blekesaune, Elvestad, and Aalberg 2012; Ksiazek, Malthouse, and Webster 2010; Toff and Kalogeropoulos 2020). Younger people’s engagement with traditional forms of news is usually characterized by both avoidance and greater distrust (Edgerly et al. 2018; Newman et al. 2022). These differences have been attributed to both life-course effects (Quintelier 2007), since younger people have fewer of those resources that are related to higher levels of news consumption at their disposal (e.g., education, income, and social ties), as well as potential generational effects, which suggest that age should not only be studied in isolation but rather be seen as a guide to belonging to a specific generational cohort (Zukin et al. 2006).

This generational perspective traces cohort differences to their shared social locations, which in turn result from common experiences of historical and social processes (Mannheim 1952). People born at similar times and places develop similar experiences, practices, and values in their youth—the most formative phase of life—and when formed, they are rather stable (Zukin et al. 2006). Media scholars have argued that these shared social practices formed early in life also shape people’s news consumption habits, suggesting that the media that were dominant during individuals’ adolescence remain central as they grow older (Ghersetti and Westlund 2018; Westlund and Weibull 2013).

Apart from generational news habits—with today's youth growing up with social and digital media—younger people’s reliance on newer news formats has also been attributed to changing generational values (Bennett 2008). Online and social media rely less on top-down one-way communication characteristics of legacy media, promoting instead a networked structure where information flows freely, open for sharing, discussing, and remixing, which have been thought to be more accessible to newer forms of participation and citizenship norms (Bennett 2008; Zukin et al. 2006). Additionally, as media choices become increasingly individualized, younger people might develop weaker relationships and habits around news consumption, at least regarding traditional news media (Edgerly 2017; Edgerly et al. 2018).

In sum, while it is reasonable to assume that younger people have always had lower levels of news consumption due to life-course effects (Quintelier 2007), we can expect that this age gap has increased as the youngest generations find themselves growing up navigating an increasingly complex high-choice media environment. Thus:

H2a: Younger people are more likely to avoid news media; and

H2b: The size of the relationship between younger age and news media avoidance has increased over time.

Integrating the Motivational and Generational Perspectives

Perhaps the clearest gap in the literature on news avoidance is that little has been done to integrate the motivational and generational perspectives as we have transitioned from low-choice to high-choice media environments. The motivational perspective has traditionally overlooked generational effects, arguing that “people have not necessarily changed; they have merely changed the channel” (Prior 2007, 19). As for the generational perspective, much focus has been on news habits. While news habits are important (Edgerly et al. 2018; Ghersetti and Westlund 2018), we should also consider the role of generational values in shaping motivations for, or avoidance of, news consumption. This lack of inquiry into the relationships between a changing media environment, changing audiences, and changing relationship to news is unfortunate since these processes are inextricably linked.

Theoretically, scholars have suggested a shift in citizenship norms, where older generations are oriented towards institutionalized dutiful citizenship norms, while younger generations are increasingly characterized by self-actualizing citizenship norms driven by personal motivations (Bennett 2008; Dalton 2008). In Thorson’s (2015) interviews with adolescents, she found that the most widely shared norm about a good citizen was that any specific expression of political participation, such as following the news, was optional—an individual choice, driven by personal interest (Thorson 2015, 17). What this “do-it-yourself” citizenship norm suggests is that individual motivations are increasingly a key resource in political participation and engagement. Thorson (2015) argued that this norm is based on two trends. First, the move away from traditional membership society—characterized by group loyalties and routine attention to the news media—means that there are fewer pathways for those initially unengaged to foster a political interest. Second, if political engagement is increasingly viewed as an optional lifestyle choice, there might be a weaker social stigma for disengaging from the political world.

The suggestion that older generations to a greater degree consider news consumption as a civic duty can have implications for news avoidance and audience fragmentation. A civic duty to stay informed has also been conceptualized as an individual-level motivation within the OMA framework (Delli Carpini and Keeter 1996; Prior 2007) and linked to higher levels of news consumption (Poindexter and McCombs 2001; Trilling and Schoenbach 2013). In their comparative study, Toff and Kalogeropoulos (2020, 370) found that the stronger “cultures of news consumption” (370; i.e., norms about the value of news as a civic duty in a country), the lower the number of news avoiders. In Palmer and Toff’s (2020) interviews with news avoiders in the United Kingdom and Spain, a recurring theme among the respondents was that they lacked a sense of civic duty, rejecting the notion that staying informed was required to be a good citizen.

While a sense of civic duty and political interest have both been conceptualized as individual-level motivations (Delli Carpini and Keeter 1996), they differ in the sense that political interest is a preference-based motivation while civic duty is connected to normative concerns about good citizenship (Poindexter and McCombs 2001). This suggests that strong norms about citizens’ duty to stay informed can counterbalance preference-based fragmentation because people “may not like news as much as entertainment, but they still follow it because they consider it their duty as citizens to be informed” (Prior 2007, 261).

Empirically, the relationship between generational shifts in civic duty norms and news consumption has received less attention. Bouilanne and Shehata (2022) found that political interest was a stronger predictor of online news consumption and online political expression among younger people in the United States, United Kingdom, and France. These same results were found in Andersson’s (2019) study in Sweden: The younger a person was, the bigger the impact of political interest on news consumption. Additionally, Andersson found that among older people the gaps in news use between those with low and high political interest were smaller. However, both studies (Andersson 2019; Boulianne and Shehata 2022) are limited by their single cross-sectional design, making it challenging to discern whether we are dealing with life-course effects or generational shifts. A longitudinal perspective investigating both life-course and cohort differences in news use and engagement was conducted by Andersen & colleagues (2021). In their study they found that while younger generations benefited the most from political media use when it came to mobilization effects, higher political engagement was a more important precondition for exposure on both traditional and social media, suggesting a dynamic where “the rich-get-richer” (2021, 130–132). However, their panel data only spans two years which makes it difficult to distinguish between life-course and cohort differences.

This discussion suggests that political interest, a preference-based motivation, should be a stronger predictor for news media avoidance among younger generations compared to older generations. The combination of adolescents growing up in a high-choice media environment—where they might develop weaker news habits due to the ease of tuning out (Edgerly 2017; Edgerly et al. 2018)—and increasingly influenced by self-actualizing citizenship norms, the choice to engage with news media should be guided more by personal interest (Thorson 2015). Concurrently, political interest should be a less decisive factor for older generations who are both motivated to a greater extent by a sense of civic duty and were socialized into media habits formed in a low-choice media environment (Bennett 2008). Previous studies seem to support this dynamic (Andersen et al. 2021; Andersson 2019; Boulianne and Shehata 2022) but the research still lacks truly long-term research. This is an important gap to fill since we should not only expect that the impact of political interest is greater among younger people but also that this difference has increased over time. Hence, we expect that:

H3: The impact of political interest on the probability of avoiding news media has increased more among younger people than among older people over time.

Method & Data

Empirically, this study focuses on Sweden. There are three reasons for this. First, Sweden’s media system has been characterized as a typical democratic corporatist system with traditionally high levels of newspaper readership, well-financed and popular public service outlets as well as high degrees of organization and independence of journalists (Hallin and Mancini 2004). Trust in, and average use of, news media among Swedes have been high and remain stable (Newman et al. 2022; Strömbäck, Djerf-Pierre, and Shehata 2013). Hence, the Swedish context has been described as a “culture of news consumption” (Toff and Kalogeropoulos 2020, 367) and should provide a tougher test for our hypotheses. Second, in recent decades, the Swedish media environment has developed from a low-choice into a high-choice media environment (Weibull and Wadbring 2014). In the mid-1980s, Sweden still had a public service monopoly on TV and radio broadcasting. Commercial TV entered Swedish households at the end of the 80s, and commercial radio was introduced at the beginning of the 90s. In the mid-90s the Internet became widely available, and today Sweden has one of the highest Internet penetrations in the world (Newman et al. 2022). Third, Sweden has unique time-series data on audiences and news media use spanning 35 years, allowing us to track changes during the transition from a low- to a high-choice media environment.

Data

This study utilizes data from the Society, Opinion and Media (SOM)-Institute cumulative dataset, which consists of yearly cross-sectional surveys collected via postal and web questionnaires in Sweden between the years 1986 and 2021.¹ The surveys use systematic probability sampling and are nationally representative. The response rates for the SOM surveys have been traditionally high, but have slowly declined since the 2000s, settling at around 50% for the latest surveys. Still, the accuracy of measurements in the SOM surveys remain high (SOM-institute 2023).

Dependent Variable: News Media Avoidance

News media avoidance is operationalized using Strömbäck et al.’s (2013) approach by first creating an additive news consumption index and then classifying avoiders using a cut-off point based on the index mean. The media measures constituting the index must cover the entire period while at the same time being able to reflect and incorporate changes in the media environment into the existing time series. Hence, a platform-neutral but media-type-specific method was used. This means that when digital options become available, they are incorporated into the news consumption measures for that media type. Thus, no distinction is made regarding how individuals access specific forms of news, and the focus is on the frequency of consumption. For example, a respondent reading printed morning newspapers 1–2 days per week (a score of 2) and daily online (a score of 5), would score 5 on the newspaper variable. The news media variables include the following:

Newspapers: Respondents were asked how often they read at least one or more national, regional or local morning newspapers in a typical week, either in print or online, with the highest value of these determining each respondent’s score on the Newspaper consumption variable. The responses were coded: 0 = Never, 1 = More seldom, 2 = 1–2 days/week, 3 = 3–4 days/week, 4 = 5–6 days/week, and 5 = Daily.

TV news: Respondents were asked how often they watched news programs on public service TV (Rapport and/or Aktuellt) or commercial TV (TV4 Nyheterna) or took part of news from the public service TV online, with the highest value of these determining respondent’s score on the TV news variable. The responses were coded the same as Newspapers.

Radio news: People were asked how often they listened to national public service radio news (Ekot) or took part of news from them online with the highest value of these determining each respondent's score on the radio news variable. The responses were coded the same as above.

Tabloids: Respondents answered how often they read the tabloids Aftonbladet, Expressen, GT and/or Kvällsposten in print or online, with the highest value of these determining each respondent’s score on the tabloid variable. The measurement scale was changed in 2014, and the responses were coded as follows: 1986–2013: 0 = Never, 1 = More seldom, 2 = 1–2 days/week, 3 = 3–5 days/week, and 4 = 6–7 days/week. 2014–2021: 0 = Never, 1 = More seldom, 2 = 1–2 days/week, 3 = 3–4 days/week, 4 = 5–6 days/week, and 5 = Daily.

News consumption index: The index was created by combining the scores for each media type (TV news + Radio news + Newspapers + Tabloids). Values were standardized to ensure comparability and that all measures were on an interval scale, resulting in each media variable ranging between 0 (Never) to 1 (Daily or 6–7 days/week). Thus, the news consumption index ranges between 0 (no news) to 4 (all news media daily). The news consumption index (M = 2.3, SD = 0.82) shows a normal distribution (Appendix A).

News media avoidance: The dependent variable was created as a binary variable. Respondents were coded as news media avoiders if they scored equal to or less than 1 on the news consumption index, which represents roughly one and a half standard deviations below the index mean.

While categorizing news media avoiders based on a cut-off point has been used in previous studies (Strömbäck, Djerf-Pierre, and Shehata 2013; Trilling and Schoenbach 2013), the use of one is always to some extent arbitrary. Using a cut-off point based on an additive index might also discriminate against people who rely on single outlets. To avoid this, respondents who scored 1 (daily consumption) on any of the individual media measures are not classified as news media avoiders. The problem of single outlet users should, however, not be overstated. People seldom rely on single outlets, especially in high-choice media environments (Castro et al. 2022). Additionally, using a cut-off point allows us to better understand the characteristics of people with habitually low levels of news media consumption and study news consumption gaps (Palmer, Toff, and Nielsen 2023; Skovsgaard and Andersen 2020). We also created two additional news media avoidance variables representing different cut-off points, representing one standard deviation and two standard deviations below the mean respectively (Appendix B).

Independent Variables

The independent variables are political interest, age, gender, education, and time.² Time was constructed from the year each survey was conducted, ranging from 0 (the year 1986) to 35 (the latest survey in 2021). Gender was coded into a dummy variable, with women being the reference category. Education was measured using a variable consisting of three categories: 1 = low (primary education), 2 = medium (more than primary education but not college or university), and 3 = high (college or university education). Age was measured on a continuous variable between 16 and 75. Political interest was measured using an inverted 4-point scale ranging from 0 (Very interested), 1 (quite interested), 2 (hardly interested) to 3 (Not at all interested).

An additional binary political interest variable was coded, where 1 = uninterested (not at all interested + hardly interested) and 0 = interested (quite interested + very interested). This was done because the original ordinal variable produced model misspecification problems when testing H3. Collapsing the variable into two categories of interested and uninterested solved the problem, possibly due to a decrease in model complexity.

Data Analysis

To reiterate, we are interested in the impact of political interest on the likelihood of news media avoidance among citizens of different ages and over time. In other words, we want to investigate probability differences between groups. Thus, our analysis plan followed the following steps. First, since our dependent variable is binary, we conducted multivariate logistic regressions with both two-way and three-way interactions. This does however only produce coefficients in the form of odds ratios which are hard to interpret—especially in interaction models—and cannot immediately be used to determine probability differences. Thus, as a second step, we take the logistic output and calculate meaningful probabilities of news media avoidance for different levels of political interest, ages, and years using Stata 18’s margins command (Brambor, Clark, and Golder 2006). Finally, we visualize these probabilities using Stata’s marginsplot command to facilitate interpretation.

Additionally, using cumulative data consisting of repeated cross-sectional surveys means we are dealing with individuals grouped within years. Therefore, the standard errors are clustered by each year in the dataset.

Results

We begin by outlining the changes in news media use between the years 1986 and 2021. These changes are reflected in Figure 1, which illustrates the percentage of people who use different types of news media at least 5 times a week.

Figure 1. Over time changes in news media use by media type, 1986–2021.

Note: TV news (watching commercial or PSB television news or PSB news online); Newspaper print/online (reading at least one newspaper print or online); Tabloids (reading at least one tabloid print or online); Radio news (consuming PSB radio news, broadcast or online).

The most dramatic change concerns newspaper readership. It went from being the most used media type in 1986—with 80% reading the newspaper at least 5 days/week—to displaying a continuous declining trend. Newspaper consumption online was added to the SOM surveys in 2007, and while it dampened the fall, it has yet to compensate for the lost readership. Print-only newspaper reading has plummeted. In 2021 only around 30% of respondents stated that they read printed newspapers at least 5 days a week.

Television news consumption showed a decreasing trend initially but rebounded with the introduction of the commercial channel TV4 in the 90s, remaining stable thereafter. Television is currently the most popular form of news media, with approximately 60% using it at least 5 days per week.

Tabloids had declining readership until the latter half of the 90s. Survey questions regarding online tabloid consumption were included in 1998 and stopped the negative trend. In 2014 the measurement scale for tabloids was changed, which explains the sharp drop during that time.³ By 2021, the levels of readership were roughly at the same level as when measurement began.

Radio news has shown little variation over time. Except for some fluctuations in the 90s—probably due to the introduction of commercial radio—the share of people who regularly consume radio news has been stable. In recent years the consumption of radio news has increased. This is probably due to the addition of questions regarding online consumption of radio news from 2014 onwards.

The trends in Figure 1 are similar to those in Strömbäck, Djerf-Pierre, and Shehata (2013). Given that this article shares the same data (although this study contains data for an additional decade) and similarities in operationalizations, the results speak to the robustness of these trends in the Swedish media landscape. Furthermore, this study shows that the trends identified by Strömbäck, Djerf-Pierre, and Shehata (2013) have continued and, in some cases, even accelerated after that point.

Moving to news media avoidance, in Figure 2 we see that the group with very low news media use has increased in size between 1986 and 2021. In 1986, around 2% of respondents could be categorized as news media avoiders. Three-and-a-half decades later, this group makes up roughly 12% of the population.⁴

Figure 2. Longitudinal changes in percentage of news media avoiders (with 95% confidence intervals).

Note: News media avoiders = people scoring ≤ 1 on the news consumption index. See method section for details. Grey area represents 95% confidence intervals.

Motivational, Age, and Time Effects on News Media Avoidance

Turning to our hypotheses, H1a and H2a posited that low political interest and low age would increase the probability of news media avoidance. Investigating this, multiple logistic regression models using news media avoidance as the dependent variable were conducted. The results of the logistic regressions can be found in Table 1 and presented as exponentiated coefficients, also called odds ratios.

Table 1. Effects on news media avoidance 1986–2021. Logistic regression tables (exponentiated coefficients, Clustered SE).

	News media avoidance
	Model 1	Model 2	Model 3	Model 4
Political interest (ref: Very interested)
Quite interested	1.238***(0.065)	0.665 (0.147)	1.238***(0.065)
Hardly interested	2.396***(0.131)	1.282 (0.301)	2.402***(0.133)
Not at all interested	6.261***(0.411)	3.968***(1.091)	6.341***(0.421)
Time (0–35, 0 = 1986)	1.103***(0.004)	1.083***(0.008)	1.126***(0.008)	1.135***(0.009)
Age (16–75)	0.948***(0.001)	0.948***(0.001)	0.964***(0.004)	0.966***(0.006)
Education (ref: low)
Medium	0.614***(0.024)	0.614***(0.024)	0.629***(0.024)	0.595***(0.023)
High	0.604***(0.022)	0.603***(0.022)	0.617***(0.023)	0.548***(0.022)
Gender (ref: female)
Male	1.112**(0.040)	1.112**(0.040)	1.113**(0.040)	1.086*(0.040)
Political interest × time (ref: very interested × time)
Quite interested × time		1.021**(0.008)
Hardly interested × time		1.022*(0.009)
Not at all interested × time		1.015 (0.011)
Time × age			0.999***(0.000)
Political interest (binary) (ref: interested)
Uninterested				4.198***(0.939)
Political interest (binary) × time × age (ref: interested × time × age)
Uninterested × time × age				1.001*(0.000)
Intercept	0.043*** (0.005)	0.075*** (0.017)	0.024*** (0.005)	0.029*** (0.007)
Observations	141973	141973	141973	141973
Pseudo R^2	0.179	0.179	0.179	0.166

Note: Exponentiated coefficients, Clustered standard errors in parentheses. * p < 0.05, ** p < 0.01, *** p < 0.001. Education: Low (primary education), Medium (more than primary but not university/college), High (university/college). Two-way interactions included in model 4 but not shown. Political interest binary: interested (very interested + quite interested) uninterested (hardly interested + not at all interested).

The results of Model 1 (Table 1) show that the lower the political interest, the higher the odds of news media avoidance, with those not at all interested in politics being roughly 6.3 times more likely to be news media avoiders compared to those very interested in politics. Additionally, being younger is correlated with a higher probability of news media avoidance since the odds of news media avoidance decrease for each year older a person is. These patterns remain when we control for education (educated people are less likely to avoid news media), time (the probability of news media avoidance increases over time), as well as gender. In sum, both political interest and age are important factors when it comes to news media avoidance. These results support H1a and H2a.

To investigate H1b and H2b (i.e., the impact of political interest and age has increased over time), two-way interaction terms measuring the conditional relationship between political interest and time (Table 1, Model 2), as well as age and time (Table 1, Model 3) were added. Starting with the simpler interaction between age and time, in Model 3 and Figure 3(b) we can see that the impact of age on the probability of news media avoidance has increased over time. At the beginning of the time series, the effect of age was very small. Today, aging 1 year decreases the probability of news media avoidance by -.6%, which translates to a full effect of roughly a 35.5% decrease in probability. Thus, H2b is supported.

Figure 3. Longitudinal developments in (3a) Predictive margins of news media avoidance for different levels of political interest and (3b) Average marginal effect of age on probability of news media avoidance.

Note: Based on two-way interactions (3a = political interest × time; 3b = age × time) in Models 2 and 3 (Table 1); dependent variable: news media avoidance (0–1; 1 = news media avoider).

Moving to the impact of political interest over time. Model 2 shows that, compared to those very interested in politics, the quite interested and hardly interested show statistically significant increases in odds of news media avoidance over time. Interestingly the interaction term was not significant for the not at all interested. However, the significance of the interaction term alone should not determine whether there exists a meaningful conditional effect—especially when dealing with odds ratios in nonlinear regression models—and best practice is to calculate marginal effects at meaningful levels of the predictor variables (Ai and Norton 2003; Brambor, Clark, and Golder 2006; Wasserstein, Schirm, and Lazar 2019). Predictive margins of political interest over time were therefore plotted in Figure 3(a) where we see that the effect of political interest has increased for all levels of political interest over time and that the gaps have increased.

Unpacking these—somewhat unintuitive—results we begin at the top of Model 2 (Table 1), noting that the only significant difference in 1986 (i.e., when the time variable = 0) was between those very interested and not at all interested, the latter being four times more likely to be news media avoiders. Translating the odds to probabilities in Figure 3(a), they translate to very low impact—the very interested having roughly 0.6 percent probability, the not at all interested around 2.3 percent. Now, the interaction term only calculates differences in odds ratios over time, meaning that the gaps and impact may increase but if the odds ratios stay the same the interaction would be non-significant. In 2021, the probability of news media avoidance for those very interested is 8.5 and 34 percent for the least interested, a four times higher likelihood. These similarities in odds should explain the non-significant interaction term. Still, even though the odds ratios might have stayed the same, the impact of those odds has increased. A four times higher likelihood is more impactful when going from 8.5 to 34 percent than when going from 0.6 to 2.3. To substantiate this interpretation, we conducted ordinal least squares (OLS) regression which more specifically looks at group differences rather than odds ratios between groups (Appendix O). All interaction terms are highly significant and in the expected direction, supporting H1b. Lastly, we conducted pairwise comparisons of probabilities between those very interested and those not at all interested in politics (Appendix P). The comparisons show that these two groups are statistically different in the probability of news media avoidance and that the gap between these groups has increased over time. In sum, H1b is supported, the relationship between low political interest on probability of news media avoidance has increased over time.

Continuing with H3, this hypothesis posited that the impact of political interest on the probability of avoiding news media has increased more for younger people compared to older individuals. Investigating this, a three-way interaction between political interest, age, and time was added into Model 4 (Table 1). Using the original ordinal political interest variable did, however, produce model misspecification errors, which could be attributed to both the complexity of the model as well as the low numbers of politically uninterested in the youngest age groups. The choice was therefore made to collapse the political interest variable into two categories, the interested and the uninterested (see method section). This removed the misspecification errors and was therefore deemed the most appropriate model.⁵ Recoding the variable meant that we lost some explanatory power of the model, indicated by a 1.3 percentage point drop in pseudo R2. This choice was preferable though, since what we might lose in nuance, we gain in validity.⁶

The interaction term was statistically significant and to facilitate substantial interpretation the average marginal effect of political interest is presented in Figure 4. The results show that political interest is becoming more important over time across ages, as the slope increases for every year presented, placing itself higher on the y-axis. Following political interest’s marginal effect along the x-axis shows that in 1986, the effect of political interest on the probability of avoiding news was almost indistinguishable across ages, indicated by the relatively straight slope. Moving forward in time, the results show a gradual increase in the downward slope, culminating in a strong conditional relationship between age and political interest in the later periods. In 2021, the average marginal effect of being uninterested in politics on the probability of avoiding news is around 15–20% for the younger audience, while it hovers between 3–5% for older individuals.

Figure 4. Average marginal effect of being uninterested in politics on probability of news media avoidance.

Note: Based on three-way interaction (political interest × age × time) in Model 4 (Table 1); dependent variable: news media avoidance (0–1, 1 = news media avoider).

Additionally, predictive margins for news media avoidance were calculated for the politically interested and uninterested, at different ages and years (Figure 5). The predictive margins substantiate the earlier findings. There were almost no age differences in the impact of political interest at the beginning of the period, but over time a clear relationship between age and impact of political interest emerges. In 2021, younger people who are uninterested in politics are roughly 20% more likely to avoid the news media compared to their interested counterparts. For older audiences, this difference is much smaller. In sum, H3 is supported. The effect of political interest on news media avoidance has increased more for younger individuals compared to older.⁷

Figure 5. Predictions of news media avoidance for different years, ages, and levels of political interest.

Note: Grey line = uninterested in politics vs. black line = interested in politics; Based on the three-way interaction (political interest × age × time) in Model 4 (Table 1); dependent variable: news media avoidance (0–1, 1 = news media avoider).

Complementary Analyses

Although the aim of this study has been to study news media avoidance over time, a general question could be whether the results reported would change with the inclusion of news consumption on social media. Addressing this, two additional measures of news avoidance were constructed using a general question of news consumption on social media first introduced in the 2014 survey (Appendix B). The results of these complementary analyses show, on the one hand, that both the levels of news avoiders and the relationship between political interest and age on probability of news avoidance weakened when including news consumption on social media. On the other hand, the results still show that lower levels of political interest are a greater predictor of news avoidance for younger compared to older individuals, even when including news consumption on social media (see Appendix H & I).

Discussion & Conclusions

This study investigated the changing impact of political interest and age on news media avoidance in Sweden between the years 1986 and 2021, both individually and their conditional relationship. It did so by departing from research building on the OMA framework and generational changes in citizenship norms, predicting that motivational factors are not only becoming more important over time but also that these effects are stronger for younger people. The results show that being uninterested in politics and being younger increases the probability of avoiding news media. Additionally, the impact of these two factors has increased over time. Finally, this paper demonstrates that the impact of political interest on the probability of news media avoidance has increased more for younger people compared to older individuals over time.

This paper contributes to news avoidance research in multiple ways. On a surface level, the findings seem to support the arguments of the OMA framework, which states that the impact of individual-level motivations will become a more important predictor for news consumption as opportunities for media content selection increase (Luskin 1990; Prior 2007). Conceptualizing political interest as an individual-level motivation, this study has shown that the impact of this factor on the probability of news media avoidance has increased over time. The longitudinal perspective utilized in this study provides robust evidence to previous literature that has investigated the relationship between political interest and news consumption (Blekesaune, Elvestad, and Aalberg 2012; Gorski and Thomas 2021; Ksiazek, Malthouse, and Webster 2010; Strömbäck, Djerf-Pierre, and Shehata 2013). Attention should be turned to Strömbäck et al.’s (2013) article, which this study could be said to be a continuation of—both using longitudinal SOM-data and sharing similarities in operationalization. Adding a decade of data and employing a news avoidance framework, this study has shown that the trends identified by Strömbäck, Djerf-Pierre, and Shehata (2013) continue to this day, and in some ways seem to have accelerated. In sum, the findings imply that the high-choice media environment does facilitate a preference-based fragmentation of audiences (Prior 2007), at least regarding engagement with traditional news media.

At the same time, the results of this study challenge core parts of the OMA framework, which states that greater choice opportunities should translate into a stronger impact of individual motivations. As Prior (2007, 19) argued: “People have not necessarily changed; they have merely changed the channel”. By showing that low political interest is a greater predictor of news media avoidance for younger people compared to older—and that this conditionality has not always existed but rather has emerged over time—this study suggests that people have changed, at least when it comes to their relationship with news media consumption. As such, this study is in line with previous literature (Andersson 2019; Boulianne and Shehata 2022) which shows that “political interest is clearly more important among some groups of citizens than others – even when their media environment is the same” (Boulianne and Shehata 2022, 778). The fact that these dynamics seem to operate similarly both when studying traditional news media—the aim of this study—as well as for news in online and social media (Andersen et al. 2021; Boulianne and Shehata 2022) contributes to a deeper understanding of news consumption and avoidance in the high-choice media environment. These findings indicate that we are not only dealing with life-course effects or changes in media environments, growing differences strongly related to age are also taking place among the news audience.

Parts of these differences certainly reflect news habits of older cohorts socialized early in life (Ghersetti and Westlund 2018; Westlund and Weibull 2013). However, the greater—and seemingly growing—impact of political interest on news consumption among younger people suggests that more than habits are at play here. It could also reflect different norms around citizenship between younger and older generations, as suggested by some scholars (Bennett 2008; Dalton 2008), where younger generations are less willing to “subscribe to the notion held by earlier generations that citizenship is a matter of duty” (Bennett 2008, 14). These changing norms may translate to news consumption and staying informed about politics being increasingly viewed as optional, a “choice left up to the individual. And it is a choice driven by personal interest” (Thorson 2015, 17). The longitudinal and generational perspective utilized in this study does seem to support these notions. Moving forward, researchers should be conscious of the results presented in this study, that political interest is a quite conditional factor, at least when it comes to news media avoidance.

Although the aim of this study has been to study news media avoidance (i.e., low consumption of traditional news media) it is worthwhile to reflect on the role of social media on our results. As our complementary analyses shows (Appendix B, H & I), the share of news avoiders decreases and the general associations between political interest, age and news avoidance weakens—but does not disappear fully—when incorporating news use on social media. On the one hand, this supports the argument that social media may have a levelling effect on news consumption gaps by facilitating incidental exposure to news and lowering thresholds for consumption (Holt et al. 2013; Karlsen, Beyer, and Steen-Johnsen 2020). On the other hand, most findings to date lean towards the notion that consuming news from social media does not compensate for not using traditional media when learning about politics and current events (Amsalem and Zoizner 2023; Shehata and Strömbäck 2021). Additionally, we were only able to incorporate a question of general news consumption on social media while studies using more specific measures, such as political uses of social media, tend to find similar gaps as those in traditional news media (Andersen et al. 2021; Boulianne and Shehata 2022; Kümpel 2020). In sum, while social media might contribute to smaller news consumption gaps in the aggregate, we would argue caution to overestimate their de facto levelling potential. This should be especially pertinent given this article’s findings; that traditional news media still matters to younger people but—compared to older individuals—consumption is increasingly tied to political interest.

Some study limitations should be noted. First, previous research shows that survey respondents overestimate their news consumption (Prior 2009) and this is probably the case in this study as well. Hence, this study will have likely underestimated the actual levels of news media avoidance. Second, younger people and people with lower political interest are underrepresented in the SOM-surveys (SOM-institute 2023). This implies that the greater importance of political interest on news media avoidance among younger people found in this study is likely to be understated. More importantly, this is a single-country study, which implies that the generalizability of findings is unclear. Additionally, this study did not directly measure civic duty norms and/or generational effects. Instead, this study built on arguments made by previous research and theories on generational civic duty norms (Bennett 2008; Thorson 2015). The result that political interest is more important for younger people is valuable in and of itself and supports the argument that political engagement is increasingly viewed as a choice driven by personal interest rather than duty. But without controlling what citizenship norms people subscribe to, we cannot be certain. Future research should incorporate measures and methods that better capture these theorized changes in citizenship norms across generations and how they relate to news consumption. A fruitful way forward to fully investigate these independent effects of age, generation and time would be age-period-cohort (APC) analyses.

These limitations notwithstanding, if political uninterest is a stronger predictor of news media avoidance among younger people, we can expect that the preference-based gaps reported in this and earlier studies (Prior 2007; Strömbäck, Djerf-Pierre, and Shehata 2013) will continue to increase and perhaps accelerate as newer generations with weaker news habits and norms about news media consumption replace older generations. The implications of these findings also extend to other forms of political participation, since news consumption is positively related to political knowledge gains and political engagement (Andersen et al. 2021; Damstra et al. 2023; Ksiazek, Malthouse, and Webster 2010; Prior 2007). Increasing audience fragmentation gaps along preferential lines might therefore presage growing inequalities in other forms of political involvement. In such a media environment, increasingly characterized by heterogenous socialization of news habits (Aalberg, Blekesaune, and Elvestad 2013, 299), scholars should focus on how the less interested can retain a pathway into engagement with news media.

Disclosure Statement

No potential conflict of interest was reported by the author(s).

Data availability statement

The data that support the findings of this study are available from the Swedish national data service (SND). Restrictions apply to the availability of these data, which were used with permission for this study. Data are available from https://snd.gu.se/en/catalogue/dataset/snd0905-1 with the permission of the SOM institute.

Notes

1 SOM-Institute (2023). The SOM Institute Cumulative Dataset 1986–2021 v2022.1. University of Gothenburg.

2 Appendix C contains summary statistics.

3 Appendix M contains question wordings and response scales of news consumption measures. Online measures of tabloids were added before the other news media types because they established an online presence early. The later implementation of online questions for the other media types might have underestimated the usage of these outlets. Inspecting Appendix M however, the question wordings have historically been neutral with regards to way of access and could be argued to capture both offline and online news consumption during the period when specific online questions were not included. This assertion is supported in Figure 1, showing consistent trend lines without significant year-to-year fluctuation.

4 Appendix B contains a version of Figure 2 with four alternative measures of news avoidance, two of which contains a social media news use measure from 2014 and onward. The combined picture of all measures suggests that news avoidance is a growing phenomenon. However, incorporating social media news use decreases the total amount of news avoiders.

5 See appendix D for model misspecifications tests.

6 Appendix F contains analyses using the ordinal political interest variable. Please note that the ordinal model does contain misspecification errors.

7 See online Appendix B for robustness checks using alternative codings of news media avoidance.